1920 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, …haosu-robotics.github.io/images/Paper/2015TmechProstateRobot.pdf · 1922 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20,

1920 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

Piezoelectrically Actuated Robotic Systemfor MRI-Guided Prostate Percutaneous Therapy

Hao Su, Member, IEEE, Weijian Shang, Student Member, IEEE, Gregory Cole, Member, IEEE,Gang Li, Student Member, IEEE, Kevin Harrington, Alexander Camilo, Junichi Tokuda, Clare M. Tempany,

Nobuhiko Hata, and Gregory S. Fischer, Member, IEEE

Abstract—This paper presents a fully actuated robotic systemfor percutaneous prostate therapy under continuously acquiredlive magnetic resonance imaging (MRI) guidance. The system iscomposed of modular hardware and software to support the sur-gical workflow of intraoperative MRI-guided surgical procedures.We present the development of a 6-degree-of-freedom (DOF) nee-dle placement robot for transperineal prostate interventions. Therobot consists of a 3-DOF needle driver module and a 3-DOFCartesian motion module. The needle driver provides needlecannula translation and rotation (2-DOF) and stylet translation(1-DOF). A custom robot controller consisting of multiple piezo-electric motor drivers provides precision closed-loop control ofpiezoelectric motors and enables simultaneous robot motion andMR imaging. The developed modular robot control interface soft-ware performs image-based registration, kinematics calculation,and exchanges robot commands and coordinates between the nav-igation software and the robot controller with a new implementa-tion of the open network communication protocol OpenIGTLink.Comprehensive compatibility of the robot is evaluated inside a 3-TMRI scanner using standard imaging sequences and the signal-to-noise ratio loss is limited to 15%. The image deterioration due tothe present and motion of robot demonstrates unobservable im-age interference. Twenty-five targeted needle placements insidegelatin phantoms utilizing an 18-gauge ceramic needle demon-strated 0.87-mm root-mean-square (RMS) error in 3-D Euclideandistance based on MRI volume segmentation of the image-guidedrobotic needle placement procedure.

Index Terms—Biopsy, brachytherapy, image-guided therapy,MRI-guided robotics, piezoelectric actuation.

Manuscript received June 5, 2014; revised October 16, 2013, May 23, 2014,and July 29, 2014; accepted September 16, 2014. Date of publication November5, 2014; date of current version August 12, 2015. This paper was presented inpart at the IEEE International Conference on Robotics and Automation, Shang-hai, China, May 9–13, 2011. Recommended by Technical Editor S. Martel.This work was supported in part by the Congressionally Directed Medical Re-search Programs Prostate Cancer Research Program New Investigator AwardW81XWH-09-1-0191, NIH P41EB015898, R01CA111288, Link FoundationFellowship, and Dr. Richard Schlesinger Award from the American Society forQuality. (Corresponding author: Hao Su.)

H. Su and G. Cole were with the Automation and Interventional MedicineRobotics Lab, Department of Mechanical Engineering, Worcester PolytechnicInstitute. They are now with Philips Research North America, Briarcliff Manor,NY 10510 USA (e-mail: [email protected]).

W. Shang, G. Li, K. Harrington, A. Camilo, and G. S. Fischer are with the Au-tomation and Interventional Medicine Robotics Lab, Department of MechanicalEngineering, Worcester Polytechnic Institute, Worcester, MA 01609 USA.

J. Tokuda, C. M. Tempany, and N. Hata are with the National Center for ImageGuided Therapy, Brigham and Women’s Hospital, Department of Radiology,Harvard Medical School, Boston, MA 02115 USA.

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TMECH.2014.2359413

Fig. 1. (Left) A radiologist reaching the perineum of the patient inside closed-bore MRI while unable to see the navigation software, and (right) a mechanicalgrid template to guide manual needle placement. Figures are adapted from [2].

I. INTRODUCTION

MAGNETIC resonance imaging (MRI) offers high-resolution tissue imaging at arbitrary orientations and

is also able to monitor therapeutic agents, surgical tools, tis-sue properties, and physiological function, which make MRIuniquely suitable for guiding, monitoring, and controlling awide array of localized interventions [1]. Nevertheless, the lim-ited space inside the bore is typically 60–70 cm in diameter and170 cm in length. The ergonomics of manual needle placementprove very difficult in the confines of the scanner bore. Fig. 1(left) shows a radiologist reaching into the center of the scan-ner bore to perform a needle placement through the perineumof a patient, and illustrates the challenging ergonomics of thismanual insertion scenario.

A. Robot-Assisted Prostate Interventions With MRI Guidance

Numerous studies [3], [4] have shown that transrectal ultra-sound (TRUS)-guided prostate biopsy, the most common ap-proach to sample suspicious prostate tissue, provides relativelylow quality images of the tissue and needles, and thus has lim-ited ability to localize brachytherapy seeds, especially if there isa shadowing effect in a dense distribution of seeds. Due to con-cerns about MRI safety and compatibility of instrumentation,MRI is not, at this point, commonplace for guiding prostatecancer procedures.

To overcome the aforementioned challenges, robotics hasbeen introduced. In the order of increased active actuationDOFs, early MRI-guided prostate interventional systems startfrom template-guided manual straight needle placement withtransperineal approach as a proof of concept [5] at Brighamand Women’s Hospital. Fig. 1 (right) shows a patient lying ona MRI scanner table with leg supports during the clinical case.Beyersdorff et al. [6] reported transrectal needle biopsies in

1083-4435 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1921

clinical studies employing 4-DOF passive alignment (rotation,angulation, and two linear translations) of a needle sleeve (In-vivo Corp., USA). Krieger et al. presented a 3-DOF passive armand a 2-DOF motorized arm to aim a needle guide for transrectalprostate biopsy [7]. Song et al. [8] developed 2-DOF motorizedsmart template guide consisting of vertical and horizontal cross-bars that are driven by ultrasonic motors. Stoianovici et al. [9]described a pneumatic stepper motor and applied it to a newgeneration of robotic prostate interventions [10]. Our previouswork presented a pneumatic servo system and sliding mode con-trol [11] which was later adapted as a parallel manipulator forposition-based teleoperation [12].

B. Robotic Actuation in the MRI Environment

Following the definition by American Society for TestingMaterials (ASTM) standard F2503-05, in 2008, the U.S. Foodand Drug Administration redefined the classifications of MRIdevices as “MR Safe,” “MR Conditional,” and “MR Unsafe,”while the term “MR Compatible” was not redefined and is notused in current standards. The latest ASTM F2503-13 clearly de-fines “MR Safe” as “composed of materials that are electricallynonconductive, nonmetallic, and nonmagnetic.” Therefore, theproposed robot, like all electromechanical systems, strives fordemonstrating “safety in the MR environment within definedconditions” and the “MR Conditional” classification. Note thatthese terms are about safety but neither image artifact nor devicefunctionality is covered. This manuscript demonstrates that theproposed robot is both safe in the MR environment in the in-tended configuration and that it does not significantly adverselyaffect image quality.

From an actuation perspective, hydraulic actuation provideslarge power output and could be potentially MR Safe, but it isnot ideal due to cavitation and fluid leakage [13]. Pneumaticactuation inherently could be designed intrinsically MR Safe.Thus, far only the transperineal access robot [9] and transrec-tal access robot [10] by Stoianovici et al. at the Johns HopkinsUniversity and the robot by Yakar et al. [14] at the Radboud Uni-versity Nijmegen Medical Center have proven to be completelyMR Safe. A major issue of actuating with pneumatic cylindersis to maintain stability, which may result in overshooting dueto the nonlinear friction force and long pneumatic transmissionline induced slow response. Yang et al. [15] identified that thepeak-to-peak amplitude of the oscillations before stabilizationat the target location ranges in 2.5–5 mm with sliding modecontrol. Our experience with a pneumatic robotic using slidingmode control [11] demonstrated 0.94 mm RMS accuracy fora single axis. Alternatively, current pneumatic stepper motorsachieve 3.3◦ [9] and 60◦ step size [16] demonstrating limitedpositioning capability.

Piezoelectric actuators can be very compact, provide submi-cron precision, and good dynamic performance without over-shooting. Compared with pneumatics, piezoelectricity haveunparalleled positioning accuracy and power density [17]. Forexample, the rotary piezoelectric motor in our robot (PiezoLegs,LR80, PiezoMotor AB, Sweden) has a 5.73 × 10−6 degree stepangle and is 23-mm diameter by 34.7 mm long, as compared

with the novel pneumatic stepper motor described in [18] witha 3.3◦ step angle and is 70 × 20 × 25 mm. However, piezo-electric motors utilizing commercially available motor driverscause unacceptable MR imaging noise (up to 40–80% signalloss) during synchronous robot motion [7]. In previous work,Fischer and Krieger et al. evaluated various types and config-urations of piezoelectric actuation options demonstrating thelimitations of commercially available piezoelectric drive sys-tems [20]. A primary contribution of this paper is “minimizing”the image artifact caused by standard motor drivers and en-able high-performance motion control through development ofa custom control system and leveraging the advantages of theseactuators through design of a piezoelectrically-actuated robot.A historical review of MRI-guided robotics and the challengesrelated to piezoelectric actuation can be found in [13], [20],and [21].

C. Contribution

Generally, previous work in MRI-guided prostate interven-tion utilizing piezoelectric actuation is limited in the followingthree aspects. First, a majority of the previously developed sys-tems are not fully actuated; thus, remains time-consuming andnecessitating moving the patient inside the scanner for imag-ing and moving the patient out of the scanner bore for the in-terventional procedure. Second, robots utilizing commerciallyavailable drivers have to interleave motion with imaging [22] toprevent electrical noise that causes 40–80% signal degradation[7], precluding real-time MR image guidance. Third, most ofthe prior research systems are either scanner-dependent (i.e.,they require specific electrical, mechanical, and/or software in-terfaces to a scanner model/vendor or scanner room, or theyare not readily transported and setup for use in an arbitraryMRI scanner), or they are platform dependent (e.g., based ona specific control interface [23]), or lack integrated navigation,communication, visualization, and hardware control.

Hence, the primary contributions of the paper are: 1) a fullyactuated 6-DOF robot capable of performing both prostatebiopsy and brachytherapy procedures with MR image guid-ance, while keeping the patient inside the MRI bore; 2) afeature-rich motion control system capable of effectively driv-ing piezoelectric motors during real time continuously acquired3-T MRI imaging; 3) a modular hardware (robotic manipulatorand the control system) and software (image processing, semi-autonomous robot control, registration, and navigation) systemthat supports network communication and is also readily ex-tendable to other clinical procedures; and 4) a vendor-neutraland scanner room-independent system. This paper demonstratesevaluation with a Philips Achieva 3-T scanner; similar re-sults have been identified with Siemens Magnetom 3-T MRIscanner [24].

II. DESIGN REQUIREMENTS AND SYSTEM ARCHITECTURE

A. System Concept and Specifications

The prostate interventional robot design aims to model theprocedure after a radiologist’s hand motion and fit between

1922 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

Fig. 2. System architecture and data flow of the robotic system. Six modulesare shown in gray block and OpenIGTLink is used to exchange control, image,and position data.

patient legs inside an MRI scanner bore. Beyond the compati-bility with the MRI environment, the following are the primarydesign considerations.

1) Workspace: The patient lies inside MRI scanner inthe semilithotomy position, and the robot accesses theprostate through the perineal wall. The typical prostate is50 mm in the lateral direction by 35 mm in the anterior-posterior (AP) direction by 40 mm in length. To cover allvolume of prostate and accommodate patient variability,the prostate is assumed to have the shape of a sphere with50-mm diameter. Thus, the required motion range is asfollows: vertical motion of 50 mm (100–150 mm abovethe scanner bed surface), lateral motion of ±25 mm fromthe center of the workspace, and needle insertion depth is150 mm to reach the back of the prostate from the skinentry point at the perineum.

2) Sterilizability: Only the plastic needle guide, collet, nut,and guide sleeve have direct contact with the needle or thepatient and are removable and sterilizable. The remainderof the robot can be draped with a sterilized plastic cover.

B. System Architecture

The system architecture comprises the following six mod-ules depicted in Fig. 2: 1) a needle placement robot inside MRIscanner; 2) a piezoelectric robot controller inside MRI scannerroom, 3) an interface box with a fiber optic media converter androuter; 4) a control computer running robot control software;5) surgical planning, navigation, and visualization software 3DSlicer; and 6) an MRI scanner and image acquisition interface.The system is designed from the ground up to be modular,readily reconfigurable, and scanner independent. The hardwarearchitecture enables scanner independence by only requiring a

grounded ac socket in the scanner room and a waveguide tubeto pass out a fiber optic cable. There is no requirement forspecific patch panel configurations, which may not have the re-quired connections (or must be customized), may allow noise topass into the scanner room, and often have filters that degrade(or completely prevent) piezoelectric motor diver signals frompassing. The control system is modular in the sense that com-munication between all modules (except the one between robotcontroller and robot itself which are adjacent) is through a net-work connection as indicated with dashed lines, which impliesthat not only each module’s computational platforms but alsoprogramming languages are not required to be identical.

Surgical navigation software 3D Slicer, serves as a user in-terface for the surgeon. The system workflow follows a pre-operative planning, fiducial frame registration, targeting, andverification procedure as presented in Section V. OpenIGTLink[25] is used to exchange control, position, and image data. Thedetails of the module design are described in the followingsections.

Communication from the control computer (inside the con-sole room) to the robot controller (inside scanner room) isthrough fiber optic Ethernet running through a readily availablewaveguide tube in the patch panel. Computer control signals andjoint positions are communicated through fiber optic conversionbetween MRI scanner room and console room, eliminating anyelectrical signal passing in/out of the scanner room and, thus, alarge source of noise that is introduced when electrical signalpasses through the patch panel or wave guide since the cablesact as antennas to induce stray RF noise.

III. MECHANISM DESIGN

This new design has been significantly improved over ourprior work [26] with increased structural rigidity, mechanicalreliability, and ease of assembly. The major construction com-ponents of the robot (e.g., motor fixture and belt tensioner)are made with fused deposition modeling (Dimension 1200es,Stratasys, Inc., USA) and polyjet 3-D printing (Objet Con-nex260, Stratasys, Inc., USA). Plastic bearings and aluminumguide rails (Igus Inc., USA) are used in the transmission mech-anisms. Optical encoders (U.S. Digital, USA) with PC5 dif-ferential line drivers that have been shown to cause no visibleMRI artifact when appropriately configured are used for positionencoding [11].

Clinical brachytherapy and biopsy needles are composed oftwo concentric shafts, an outer hollow cannula and an innersolid stylet. Eighteen-gauge needles are typically used for clin-ical prostate brachytherapy, and radioactive seeds are typicallypreloaded with spacers between them, often 5.5 mm apart.Straight needle placement typically includes three decoupledtasks: 1) moving the needle tip to the entry point with 3-DOFCartesian motion; 2) inserting the needle into the body using1-DOF translation along a straight trajectory; and 3) firing thespring loaded mechanism of the biopsy gun to harvest tissue orretracting the stylet for placing radioactive seeds. These threetasks are implemented with two modules as described in the sec-tion that follows. The described device is a unified mechanism

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1923

Fig. 3. Exploded view of the needle driver module, including a rotary axis, atranslation axis, and a collinear stylet translation axis.

capable of being configured for the two separate procedures ofbiopsy and brachytherapy.

A. Needle Driver Module

For transperineal prostate needle placement, which requires18-N force to puncture the capsule of the prostate [27], duallinear piezoelectric motors (PiezoLegs LL1011C, PiezoMotorAB, Sweden), each providing 10 N holding force and 1.5 cm/sspeed are placed in parallel to drive 1-DOF insertion motion.This mechanism design is illustrated in Fig. 3; these motors arenot clearly visible since they are placed beneath the needle driverplate. Another 1-DOF of collinear translational stylet motion isdriven by another PiezoLegs actuator to coordinate the motionwith respect to the cannula. For biopsy, the firing motion isalso implemented as rapid cannula-stylet coordinated motionusing the same coaxial mechanism. This idea unifies the twoprocedures and simplified mechanism design.

The needle driver’s design allows standard needles withdifferent diameters, ranging from 25 (0.51-mm diameter) to16 gauge (1.65-mm diameter) to be used. A collet clampingdevice rigidly couples the cannula shaft (outer needle) to thedriving motor mechanism as shown in Fig. 3. The clampingdevice is connected to the rotary motor through a timing belt. Aplastic bushing with an eccentric extruded cut is used as pulleytensioner, where the extruded cut is used to house a rotary piezo-electric motor (PiezoLegs, LR80, PiezoMotor AB, Sweden).Rotation of the belt tensioner can adjust the distance betweenthe motor shaft and the collet shaft of clamping mechanism. Inthe previous robot iteration [26], collet screw shaft bending wasobserved and the mechanism has been significantly improvedwith a newly designed tensioner that includes a shaft supportwith plain bearing to prevent motor shaft bending. The plasticneedle guide with a press-fit quick release mechanism, colletnut, and guide sleeve that have direct contact with the needleare, therefore, readily detachable and sterilizable.

B. Cartesian Motion Module

The needle driver is placed on top of a 3-DOF Cartesianpositioning module shown in Fig. 4. The mechanism design was

Fig. 4. Side view of the 3-DOF needle driver module providing cannulatranslation and rotation (2-DOF) and stylet translation (1-DOF) and the 3-DOFactuated Cartesian stage module.

developed in a decoupled manner, thus the kinematics are simpleand safe by separating the alignment and insertion motions. Thehorizontal 2-DOF motions are achieved using the same linearmotors with direct drive. Motion in the vertical plane enables40 mm of vertical travel. To increase rigidity of the verticalmechanism over the previous prototype [26], [28] that has asingle scissor mechanism, the new design utilized a one-and-half Scott-Russell scissor mechanism which is compact andattenuates structural flexibility due to plastic bars and bearingsas identified in Fig. 4. Vertical motion is achieved by actuatingthis mechanism with a aluminum anodized lead screw (2-mmpitch). The piezoelectric motors exhibit inherent braking whennot actively driven providing further safety.

The piezoelectric motors used here are capable of microm-eter level positioning accuracy, and the limiting factor is theclosed-loop position sensor. The optical encoders in this robothave encoding resolution of 0.0127 mm/count (EM1-0-500-I)and 0.072◦/count (EM1-1-1250-I) for linear and rotary motion,respectively.

IV. ELECTRICAL DESIGN

A. Piezoelectric Motors

Piezoelectric motors are one of the most commonly utilizedclasses of actuators in MRI-guided devices, as they operateon the reverse piezoelectric effect without a magnetic field asrequired by traditional motors. In terms of driving signal, piezo-electric motors fall into two main categories: harmonic andnonharmonic. Both have been demonstrated to cause interfer-ence within the scanner bore with commercially available drivesystems [7]. While these motors operate on similar basic prin-ciples, signals required to effectively utilize and control themare quite different. Harmonic motors, such as Nanomotion mo-tors (Nanomotion Ltd., Israel), are generally driven with fixedfrequency sinusoidal signal on two channels at 38–50 kHz andvelocity control is through amplitude modulation of 80–300 VRMS. Shinsei harmonic motors (Shinsei Corporation, Japan),

1924 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

however, are speed controlled through frequency modulation.Nonharmonic motors, such as PiezoLegs motors, operate at alower frequency (750–3000 Hz) than harmonic motors. Theseactuators require a complex-shaped waveform on four channelsgenerated with high precision at fixed amplitude (typically a lowvoltage of <50 V), and speed is controlled through modulatingthe drive frequency.

B. Piezoelectric Motor Driver and Control

Though there have been efforts to shield motors (such aswith RF shielding cloth) and ground the shielded control ca-bles (e.g., the most recent one from Krieger et al. [7]), the MRIcompatibly results demonstrate that signal-to-noise ratio (SNR)reduction is up to 80%. From our prior study [19], the sourceof noise is primarily from the driving signal rather than the mo-tor itself. Commercially available piezoelectric motor driverstypically use a class-D style amplification system that gener-ates the waveforms by low-pass filtering high-frequency squarewaves. By using switching drivers (the typical approach sincethe manufacturers primary design goal was power efficiency in-stead of noise reduction), significant RF emissions and noise onthe motor drive lines are introduced which cause interferencewith imaging as demonstrated in [7] and [19] and in many otherworks. Although filtering can improve the results, it has not beeneffective in eliminating the interference and often significantlydegrades motor performance.

This manuscript describes our implementation of a piezo-electric motor driver with signals generated from a direct digitalsynthesizer (DDS), high-performance multichannel digital-to-analog converter (DAC), high power linear amplifiers, and πfiltered outputs. In contrast to commercial drivers based onhigh-frequency switching voltage regulators, it is capable ofcleanly generating both high-voltage sinusoidal signals and lowvoltage precise waveforms such that it could be used to driveboth harmonic (e.g., Shinsei and Nanomotion) and nonharmonic(e.g., PiezoLegs) commercial motors [24]. A block diagram ofthe driver, shown in Fig. 5, is based on linear amplifiers withvery high speed DACs that allow precise waveform shape con-trol while eliminating any unwanted noise (both by defining asmooth waveform and through the circuit design including iso-lation, filtering, and shielding), thus fundamentally addressingthe driving signal-induced image interference.

PiezoLegs motors were the actuators of choice in this robotand, therefore, are the focus of the detailed discussion and eval-uation. They consist of four quasi-static legs (A, B, C, and Dlegs) forming a stator which interchangeably establish frictionalcontact to a ceramic drive preloaded with a beryllium copper leafspring, and are operated below their resonant frequency. Eachbimorph leg consists of two electrically isolated piezoelectricstacks. The legs elongate when an equal voltage is applied to thetwo stacks of one leg. Applying different voltages on the twostacks of one leg causes the leg to bend.

As shown in Fig. 5, the developed custom motor driverutilizes a microcontroller (PIC32MX460F512L, MicrochipTech.) and a field-programmable gate array (FPGA, CycloneEP2C8Q208C8, Altera Corp.). The functionality of the micro-

Fig. 5. Block diagram of the system components of each piezoelectric motordriver board. An FPGA is configured as a waveform synthesizer generating thefour phases of the drive signal which are passed to a linear amplifier stage withintegrated filtering.

controller includes: 1) loading waveform from SD card to theinternal RAM of FPGA; 2) communication with PC throughEthernet to receive position or velocity set points; 3) real-timeposition and velocity servo control loops. The microcontrollerincorporates a joint-level controller that controls the output sam-ple rate of the FPGA’s waveform synthesizer. The functionalityof the FPGA includes: 1) DDS; 2) encoder decoding, wheredifferential encoder signals pass through a differential receiverfor improved robustness to noise, are interpreted by the FPGA,and are fed to on-board microcontroller; 3) overheat and faultprotection. An internal temperature sensor in the motor am-plifier (OPA549, Texas Instruments, USA) is used to managethermal shutdown of the output; 4) stall and limit detection. Thesystem allows the control PC to communicate with the driverin three ways: Ethernet, universal serial bus (USB), and serialcommunication—in this application, Ethernet is used. Commu-nication with the robot controller between driver boards utilizeda serial peripheral interface (SPI) bus. A watchdog mechanismdetects the stall motion of the motor after 5-ms time out asa safety mechanism during joint limit detection and the hom-ing procedure. To drive the motor, arbitrary-shaped waveformsprovided by the manufacturer are resampled and loaded ontothe SD card. One pair of legs (A and C legs) typically has thesame waveform, while the other pair (B and D legs) uses thesame waveform with 90◦ phase shift. The DDS on the FPGAgenerates these four waveforms and the signal frequency isexpressed as

F =M

2nFclock (1)

where M is the frequency control tuning word from the micro-controller, n is the bit of phase accumulator (32 bit), and Fclockis the clock frequency (50 MHz). For a desired frequency, thecorresponding M is calculated in the FPGA. A 32-bit registerin the FPGA is used as phase accumulator (shown as blue block

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1925

in Fig. 5) to produce the digital phase, whose top 11 bits areused as an index to the 11-bit waveform table (2048 elementlong lookup table, shown as green block in Fig. 5) that maps thephase to corresponding waveform.

The generated digital waveform is converted to an analog sig-nal through a pair of two-channel high speed DACs (DAC2904,Texas Instruments, USA) shown as purple block in Fig. 5. Thedriver boards include a high power output amplification mod-ule, which passes the signals from four linear amplifiers to theactuators through high efficiency π filters to further removethe high-frequency noise. This module includes three stages oflinear amplifiers in sequence.

As shown in (2), a variation of a discrete nonlinear propor-tional integral derivative controller is implemented in micro-controller to control the motor motion. Since these piezoelectricmotors have inherent breaking, high bandwidth, and low iner-tia, damping with a derivative term is typically not required forsmooth motion with no overshoot

u =

⎧⎪⎪⎨

⎪⎪⎩

0, |e| ≤ d

kpe + ki

∫edt ⊆ [fmin , fmax ], d <e < l

umax , e ≥ l

(2)

where, u is the control signal, e is the position error, fmin andfmax are the minimum (750 Hz) and maximum (3 kHz) drivingfrequencies, respectively. kp and ki are the PI control parame-ters, although kd is implemented, it is set to zero and, therefore,left out of this equation. d and l are the minimum and maximumposition threshold values for linear frequency control. d is thedead-band and it is set to one encoder tick to eliminate chatter;it is implemented inside the FPGA running at 50 MHz and iscapable of stopping the low inertia motors at the setpoint al-most instantaneously. umax is saturation that ensures the motorto run at the specified maximum velocity associated with fmaxwhen the position error is greater than a predetermined value.To maintain high-speed motion while not damaging the mo-tor, the control frequency is limited by minimum speed (whichshould be reached at the target) and maximum speed (which itshould never exceed). kp is tuned to be large to allow the motorto run at the capped maximum speed until very close to thetarget.

Fig. 6 shows the portable shielded aluminum robot controllerenclosure which houses the custom-made piezoelectric motordriver boards configured for up to eight-axis control. A key stepin reducing image degradation from electrical noise is to encaseall electronics in a continuous Faraday cage to block as muchelectromagnetic interference being emitted from the equipmentas possible. This cage is extended through the shielded cablescarrying electrical signals out of the cage through serpentinewave guides located on the vertical sides of the control box.Standard 120 V AC power is filtered and passes to a linearregulator with 48 V DC output which provides power to thepiezoelectric motor drivers and additional regulators to generatelower voltage power for the control electronics.

Fig. 6. (Left) MRI robot controller enclosed inside a carry on travel case.(Right) Piezoelectric motor controller. Red box shows the digital control sectionwith microcontroller for closed-loop control and FPGA for waveform synthesis.Blue box shows the high speed parallel DACs and preamplifier circuits. Yellowbox shows the high power filtered linear amplifier stage.

Fig. 7. Coordinate frames of the robotic system for registration of the robotto MR image space, robot kinematics, and encoded needle insertion.

V. REGISTRATION, CONTROL SOFTWARE, AND WORKFLOW

The primary functionality of the robot control interface soft-ware developed as a part of this paper is to enforce and integratethe system workflow, including fiducial registration, robot reg-istration, kinematics, and motion control. The robot control in-terface software connects navigation software to the robot con-troller to exchange robot commands using the OpenIGTLinkprotocol. Three-dimensional surgical navigation software 3DSlicer serves as a user interface to visualize and define targetin image space; however, the modular system may be readilyadapted to other platforms.

A. Surgical Navigation Coordinates

The robot is registered to the patient coordinate system (re-ferred to as right-anterior-superior (RAS) coordinates) basedon imaging a fiducial frame (referred to as the Z-frame) at-tached to the robot as shown in Fig. 7. Details of the multislice-based fiducial registration can be found in our recent work [29];this evaluation of registration accuracy showed subpixel resolu-tion with a mean error of 0.27 mm in translation and 0.16◦ inorientation.

1926 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

After the registration phase, the robot can accept target coor-dinates represented in RAS image coordinates.

The corresponding series of homogeneous transformations isused to determine the robot’s tip location in MR image coordi-nates (i.e., RAS coordinates)

TRASTip = TRAS

Z · TZRob · TRob

Tip (3)

where TRASTip is the needle tip in the RAS patient coordinate

system, TRASZ is the fiducial’s 6-DOF coordinate in RAS co-

ordinates as determined by the Z-frame fiducial-based regis-tration, TZ

Rob is the robot needle guide location with respect tothe robot base (coincident with fiducial’s 6-DOF coordinate) asdetermined from the forward kinematics of the robot as definedin Section V-B, and TRob

Tip is the needle tip position with respectto the front face of the needle guide (typically at the skin entrypoint) of the robot as measured by optical encoder along theneedle insertion axis.

B. Robot Kinematics

With the navigation coordinate relation, we can substitutethe kinematics into the kinematic chain to calculate the needletip position and orientation. The vertical motion of the robot isprovided by actuation of the scissor mechanism. A linear motionin the superior-inferior direction produces a motion in the APdirection. The forward kinematics of the robot with respect tothe fiducial frame is defined as

x = q1 + xoffset

y =√

L2 − (L − d)2 + yoffset

z = q3 + q4 + zoffset (4)

where d = q2 ·p360 is the horizontal linear motion of the lead screw

due to the rotary motor motion and p is the lead screw pitch.q1 , q2 , q3 , q4 are joint space motion of the x-axis motor transla-tion (unit: mm), y-axis rotary motor ration (unit: degree), z-axismotor translation (unit: mm), and needle driver insertion trans-lation (unit: mm). L is the length of the scissor bar. The threeoffset terms in (4) are corresponding to the homogeneous trans-formation matrix TZ

Rob in (3). q4 is corresponding to the trans-formation TRob

Tip and the remainder of (4) is corresponding tothe transformation TZ

Rob .

C. Control Software for Communication and Kinematics

The control software is transparent to clinician and is devel-oped for communication and kinematics computation based onthe calculation from the last sections. A robot kinematics andOpenIGTLink communication library was developed in Javawith a graphical user interface (GUI) that coordinates systeminput, communication, and procedure workflow, while the clin-ician interfaces with the higher level navigation software (e.g.,3D Slicer). As shown in Fig. 8, the technical/engineering GUIfor the robot control application based upon the developed Javalibrary includes four major modules. 1) Robot registration andcalibration module (upper left column). It displays the registra-tion matrix TRAS

Z and matrix TZRob . 2) The upper middle column

Fig. 8. GUI of the robot control software running on the control computer incontrol room. The upper left column (1) is robot registration and calibration,the upper middle column (2) displays the current target point transformationand current actual robot 6-DOF pose. The upper right column (3) contains thedesired target transformation and buttons to control biopsy or brachytherapyprocedures. The bottom row (4) is the joint space panel.

displays the current target and current actual 6-DOF pose ashomogeneous transformation matrices. 3) The upper right col-umn contains a homogeneous transformation representing thedesired new target. Additional buttons are included to controlbiopsy or brachytherapy procedure steps. 4) The bottom rowis the joint space panel which is automatically generated for arobot mechanism configuration through an extensible markuplanguage (XML) configuration file containing the name of eachjoint, the actual joint position, the desired joint position, motorjogging, and home buttons.

D. Clinical Software and Workflow

On top of the control software, a clinical software modulewas developed in 3D Slicer to manage clinical workflow andcommunicate with robot and imaging workstation. A virtualneedle of the clinical software module accepts the pose infor-mation through OpenIGTLink from the robot and displays withMRI volume to assure placement safety and confirm placementaccuracy as shown in Fig. 9, where the yellow spheres indicatethe selected targets inside MRI volume in RAS coordinates.

The workflow mimics traditional TRUS-guided prostate in-sertions. The system components of the control software followthe workflow and enable the following five phases.

1) System Initialization: The hardware and software systemis initialized. In this phase, the operator prepares the robotby connecting the robot controller and attaching sterilizedneedle and needle attaching components to the robot. Therobot is calibrated to a predefined home position and loadsthe robot configuration from the XML file.

2) Planning: Preoperative MR images are loaded into 3DSlicer and targets are selected or imported.

3) Registration: A series of transverse images of the fiducialframe are acquired. Multiple images are used to performmultislice registration to enhance system accuracy.

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1927

Fig. 9. Clinical GUI integrated inside 3D Slicer shows a virtual needle whoseposition was streamed from control software was inserted to move to the selectedtargets (shown as yellow spheres) during needle placement accuracy evaluation.The needle tracks were visualized inside the MRI volume. The first column ofthe bottom subfigures shows the 5 × 5 needle tracks.

4) Targeting: Needle target is selected from 3D Slicer andthis desired position is transmitted to the robot to pro-cess inverse kinematics and the calculated joint commandis used to drive piezoelectric motors. Targets or adjust-ments may also be directly entered or adjusted in the robotcontrol interface software. Real-time MR images can beacquired during insertion that enables visualization of thetool path in MRI with the overlaid reported robot position.

5) Verification: The robot forward kinematics calculate ac-tual needle tip position (from encoder measurementsand registration results) which is displayed in 3D Slicer.Postinsertion MR images are acquired and displayed withoverlaid target and actual robot position.

VI. EXPERIMENTS AND RESULTS

To demonstrate compatibility with the MRI environment andaccuracy of the robotic system, comprehensive MRI phantomexperiments were performed. Imaging quality and needle place-ment accuracy were systematically analyzed.

A. MR Image Quality Evaluation

We evaluated compatibility of the needle placement robotin a Philips Achieva 3-T system. This evaluation includes twoassessments to quantify the image quality: 1) SNR ratio anal-ysis based on National Electrical Manufacturers Association(NEMA) standard MS1-2008; and 2) image deterioration factoranalysis based on the method in [30]. The analysis utilized a pe-riodic image quality test (PIQT) phantom (Philips, Netherlands)that has complex geometric features, including uniform cylin-drical cross section, and arch/pin section. The robot is placedin close proximity (5 mm) to the phantom. The controller wasplaced approximately 2 m away from the scanner bore insidethe scanner room. Fig. 10 (first row) shows the system setup.

The image quality analysis was performed with four imagingprotocols shown in Table I (the first three is for quantita-tive analysis with a phantom and the fourth is for qualitativeassessment with a human volunteer): 1) diagnostic imaging

Fig. 10. (First row) The robot prototype configuration in a 3-T MRI scannerwith a feature-rich PIQT phantom for the evaluation studies. (Second row)Representative T1W-FFE images of the cylindrical cross section for each of fourconfigurations. (Third row) Corresponding T1W-FFE images of the arch/pincross sections.

TABLE ISCAN PARAMETERS FOR COMPATIBILITY EVALUATION

Protocol TE TR FA Slice Bandwidth(ms) (ms) (deg) (mm) (Hz/pixel)

T1W-FFE 2.3 225 75 2 1314T2W-TSE-Planning 90 4800 90 3 239T2W-TSE-Needle 115 3030 90 3 271T2W-TSE-Prostate 104 4800 90 3 184

T1-weighted fast field echo (T1W-FFE); 2) diagnostic imag-ing T2-weighted turbo spin echo for initial surgical planningscan (T2W-TSE-Planning); 3) diagnostic imaging T2-weightedturbo spin echo for needle confirmation (T2W-TSE-Needle);and 4) a standard T2-weighted prostate imaging sequence (T2W-TSE-prostate). All sequences were acquired with field of view256 mm × 256 mm, 512 × 512 image matrix, and 0.5 mm ×0.5 mm pixel size.

Four configurations were evaluated for SNR analysis: 1) base-line of the phantom only; 2) robot present but controller androbot power off (controller power is the power from ac socketand robot power is controlled by emergency stop); 3) controllerand robot powered; and 4) motor powered on and robot is inmotion during imaging. Fig. 10 (second row) illustrates the rep-resentative images of SNR test with T1W-FFE images in thedifferent configurations.

The following imaging sets were acquired twice for eachimaging sequence in the Philips 3-T scanner to calculate theimage deterioration factors:

1) Baseline (BL1 and BL2): baseline image of the phantomwithout the robot;

1928 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

Fig. 11. Boxplot of the SNR for four robot configurations under three scanprotocols. The “*” represents the mean SNR, the horizontal line in the middleof the box represents the median, the top and bottom represent the quartiles, andthe whiskers represent the limits.

TABLE IIMEAN SNR CHANGE IN DIFFERENT CONFIGURATIONS

Protocol Baseline Robot Present Powered Motor Running(%Change) (%Change) (%Change)

T1W-FFE 23.32 22.735 (2.79%) 21.54 (7.63%) 19.86 (14.87%)T2W-TSE-Planning 106.69 104.27 (2.27%) 94.80 (11.14%) 90.96 (14.74%)T2W-TSE-Diagnosis 61.98 60.19 (2.89%) 55.65 (10.23%) 52.91 (14.64%)

2) Robot Present (RO1 and RO2): image of the phantom withthe robot in place, powered off;

3) Robot Motion (RM1 and RM2): image of the phantomwith the robot in motion.

For statistical analysis, SNR based upon the NEMA standarddefinition is utilized as the metric for evaluating MR image qual-ity. SNR was calculated as the mean signal in the center of thephantom divided by the noise outside the phantom. Mean signalis defined as the mean pixel intensity in the region of interest.The noise is defined as the average mean signal intensity in thefour corners divided by 1.25 [31]. In Fig. 11, the boxplot showsthe variation in SNR for images taken in each configuration forthe three imaging sequence. The change of SNR mean value isshown in Table II.

Fig. 12. Passive (P) and active (A) image deterioration factor ε variation dueto the present and motion of the robot.

TABLE IIIDETERIORATION FACTOR FOR PASSIVE AND ACTIVE TESTS

Protocol EP % EA %

T1W-FFE 0.000015 0.00012T2W-TSE-Planning 0.00139 0.00019T2W-TSE-Diagnosis 0.00036 0.00013

Although SNR analysis shows promising result, it cannotfully characterize the property of image deterioration. There-fore, image deterioration analysis [30] was conducted to estab-lish a quantitative image deterioration metric due to the presentof the robot in the imaging field (EP —passive test) and by therobot motion (EA —active test). As defined in [30], measuresEP ≤ 2% and EA ≤ 1% are associated with unobservable im-age interference. Utilizing the same notation [30], T notatespassive or active configuration set. The error εT 1 reflects thenormal level of noise from the imaging system itself. The fac-tors EP and EA indicate the image deterioration due to materialsin the passive test (P) and the electronics in the active test (A),respectively. Similar to Stoianovici et al. [10]:

1) εP 1 reflects differences between two sets taken withoutthe robot (BL1-BL2);

2) εP 2 reflects differences between two sets taken withoutand with the robot (BL1-RO1), a representation of thematerials-induced image deterioration;

3) εA1 reflects differences between two sets taken with therobot present and without robot motion (RO1-RO2);

4) εA2 reflects differences between two sets taken withthe robot present (no motion) and with robot motion(RO1-RM1), a representation of the electronics-inducedimage deterioration.

Fig. 12 shows the plot of one representative protocol(T2W-TSE-Diagnosis) with these parameters. The average dif-ference between εP 2 and εP 1 produces the passive image dete-rioration factor EP , and the average difference between εA2 andεA1 produces the active factor EA . Their experimental valuesare shown in Table III.

Based on the promising qualitative analysis of the robotic sys-tem, it was furthered evaluated with human prostate anatomy. Ahealthy volunteer lay in the semilithotomy position and was im-aged with T2W-TSE-Prostate protocol in Table I. Fig. 13 depicts

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1929

Fig. 13. MRI images of human prostate under different conditions: baseline,without robot (a), controller present and unpowered (b), and robot running (c).

the MR images under each of the conditions: baseline, withoutrobot, controller present and power off, and robot running.

The major observations from these two quantitative and onequalitative evaluation include:

1) the different imaging sequences demonstrated similar be-havior SNR reduction. This is also observed in [7];

2) the present of the robot with power off introduces2.79%, 2.27%, and 2.89% SNR reduction for T1W-FFE,T2W-TSE-Planning, and T2W-TSE-Diagnosis, respec-tively. This reflects the material caused SNR reduction;

3) the SNR reduction in the configuration of the robotpowered but not running (7.63%, 11.14%, and 10.23%for each sequence) is less than that the motor running(14.87%, 14.74%, and 14.64%). This reflects the electri-cal noise caused SNR reduction. In comparison, the meanSNR of baseline and robot motion during T1 imaging byKrieger et al. [7], reduced approximately from 250 to 50(80%), while it changed from 380 to 70 (72%). Our sys-tem shows significant improvement over [7] that utilizedcommercial motor drivers when the robot was in motion.As shown in Figs. 10 and 13, there is no readily visibledifference;

4) the deterioration factor shows unobservable interferencefor all protocols in all configurations tested, according tothe requirement EP ≤ 2% and EA ≤ 1% [30].

B. Phantom-Based Accuracy Evaluation of Multiple NeedlePlacements in MRI

Joint space accuracy has been evaluated in [32] demonstratingan average 0.03 mm tracking accuracy with the piezoelectricmotors. The robot was further evaluated in a phantom studyunder MR imaging. The phantom used for the experiments wasmuscle-like ballistic test media (Corbin Inc., USA). The rubber-like material was molded into a 10 cm × 10 cm × 15 cmrectangular form. This phantom was placed inside a standardflex imaging coil configuration, and the robot was initialized toits home position in front of the phantom similar to that shownin Fig. 10.

The goal of this trial was to assess the robot’s inherent ac-curacy and, therefore, the authors strove to limit paramagneticartifact of the needles, which would affect assessment of therobot’s accuracy. Clinical MRI needles are typically made of ti-tanium, which still produces a needle track artifact greater thanthe actual needle size; therefore, 18-gauge ceramic needles wereinserted into a gelatin phantom to assess robot instrument tip po-

Fig. 14. Twenty-five targets forming a semi ellipse to simulate the superiorpart of a prostate were selected from the MRI volume. Actual needle tip positiondetermined from segmented MRI image are superimposed with the 25 targets.

sition. The ceramic needle was custom made from ceramic rods(Ortech Ceramics, USA) and the tip was ground to symmetricdiamond shape with a 60◦ cone shape. In the clinical proce-dures, 18-gauge needle made of low artifact titanium (InvivoCorp, USA) are typically used.

T2W-TSE-Diagnosis is utilized to visualize the needle in-sertion trajectory. 25 needle placement targets (virtual targetsspecified in MRI volume) that form a semiellipsoid shape areselected from the MRI volume, as shown in Fig. 14. The trans-verse plane distance of the grid is 5 mm to mimic a mechanicaltemplate of TRUS procedure. Needles were inserted without us-ing the needle rotation motion to demonstrate the robot accuracyindependent of additional control methods that may further beused to improve accuracy.

Needle placement accuracy was evaluated along each direc-tion as shown in Fig. 15. The actual needle tip position is man-ually segmented from postinsertion MRI volume images. Thedesired targets and the needle tip positions were registered withpoint cloud to cloud registration. The maximum error is lessthan 1.2 mm as shown in Fig. 15 and it indicates a consistentaccuracy trend of the image-guided robotic system. The RMSerror as calculated with 3-D Euclidean distance between thedesired and actual needle tip position is 0.87 mm with a stan-dard deviation of 0.24 mm, which is on par with the MR imageresolution.

VII. DISCUSSION AND CONCLUSION

This paper presents a fully-actuated robotic system for MRI-guided prostate interventions using piezoelectric actuation. Inthis paper, we developed and evaluated an integrated modularhardware and software system to support the surgical workflowof intraoperative MRI, with percutaneous prostate interventionas an illustrative case. Extensive compatibility experiments havebeen conducted to quantitatively evaluate the SNR reduction andimage deterioration factor and qualitatively the image quality of

1930 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

Fig. 15. Needle placement errors of 25 targeted needle placements along x-y-zaxis components with the 3-D Euclidean distance RMS error 0.87 mm.

human prostate tissue. Since the latest ASTM F2503-13 stan-dard requires only fully nonconductive components to be con-sidered “MR Safe,” this system (based upon piezoelectric actu-ation which still relies on electrical currents and wires) has beendemonstrated to be “MR Conditional” which indicates that it issafe in the intended configuration. The SNR analysis and imagedeterioration analysis experiments with the complex phantomdemonstrate that this robotic system causes no major image ar-tifacts or distortion, paving for the route to regulatory clearance.Phantom experiments validated the capability of the system toexecute actuated prostate biopsy placement, and the spacial ac-curacy is 0.87 mm ± 0.24 mm. In comparison with pneumatics,Yang et al. [15] reported 2.5–5 mm-oscillations for each axiswith sliding mode control. Our pneumatic robotic system withsliding mode control [11] demonstrated 0.94-mm RMS accu-racy for single axis. The piezoelectric actuation demonstrate aclear advantage over the pneumatic counterpart. Since this mo-tor is capable of micrometer level accuracy, encoder resolutionis the effective limit. For our application, 0.03 mm single jointmotion accuracy is more than sufficient. In general, the rootcause of needle placement error is from two sources, namelythe needle-caused error and robotic system itself. Even thoughour transperineal approach with MRI guidance can avoid theTRUS probe-induced prostate motion and deformation, needle-caused error includes prostate translation, rotation, and defor-mation during needle insertion, needle deflection due to needletissue interaction, and susceptibility artifact shift of needle dueto titanium or nickel titanium material of the needle itself. Interms of the robot-related error, it includes the error from fidu-cial registration and intrinsic error of the manipulator itself dueto positioning accuracy of the motor, structure flexibility, andmisalignment. In this paper, we aim to isolate the targeting ac-curacy of the robotic system by utilizing virtual targets andceramic needles. So the error in this experiment reflects theaccumulation of errors due to the robot positioning limitation,

image-based registration, and needle deflection, excluding theerror due to susceptibility artifact shift or organ motion.

Future work would focus on addressing the related issuesincluding utilizing more rigid material (e.g., ultem and PEEKrather than ABS and acrylic), using needle rotation during inser-tion to decrease deflection, and modeling susceptibility artifact.As the 3-D position of the needle is available, we can dynami-cally adjust the scan plane of MRI to visualize needle insertionprocess, which could be used to track needle trajectory to in-crease the safety and accuracy.

ACKNOWLEDGMENT

The content is solely the responsibility of the authors anddoes not necessarily represent the official views of the PCRP,NIH or ASQ. N. Hata is a Member of the Board of Directors ofAZE Technology, Cambridge, MA, and has an equity interest inthe company. His interests were reviewed and are managed byBWH and Partners HealthCare in accordance with their conflictof interest policies.

REFERENCES

[1] F. A. Jolesz, “Intraoperative imaging in neurosurgery: Where will thefuture take us?” in Intraoperative Imaging, ser. Acta NeurochirurgicaSupplement, M. N. Pamir, V. Seifert, and T. Kiris, Eds. New York, NY,USA: Springer, 2011, pp. 21–25.

[2] J. Tokuda, K. Tuncali, I. Iordachita, S.-E. Song, A. Fedorov, S. Oguro,A. Lasso, F. M. Fennessy, C. M. Tempany, and N. Hata, “In-bore setup andsoftware for 3T MRI-guided transperineal prostate biopsy,” Phys. Med.Biol., vol. 57, pp. 5823–5840, Sep. 2012.

[3] H. S. Bassan, R. V. Patel, and M. Moallem, “A novel manipulator forpercutaneous needle insertion: Design and experimentation,” IEEE/ASMETrans. Mechatronics, vol. 14, no. 6, pp. 746–761, Dec. 2009.

[4] N. Hungr, M. Baumann, J.-A. Long, and J. Troccaz, “A 3-D ultrasoundrobotic prostate brachytherapy system with prostate motion tracking,”IEEE Trans. Robot., vol. 28, no. 6, pp. 1382–1397, Dec. 2012.

[5] A. V. D’Amico, C. M. Tempany, R. Cormack, N. Hata, M. Jinzaki,K. Tuncali, M. Weinstein, and J. P. Richie, “Transperineal magneticresonance image guided prostate biopsy,” J. Urology, vol. 164, no. 2,pp. 385–387, Aug. 2000.

[6] D. Beyersdorff, A. Winkel, B. Hamm, S. Lenk, S. A. Loening, andM. Taupitz, “MR imaging-guided prostate biopsy with a closed MR unitat 1.5 T: Initial results,” Radiology, vol. 234, no. 2, pp. 576–581, Feb.2005.

[7] A. Krieger, S. Song, N. Bongjoon Cho, I. Iordachita, P. Guion,G. Fichtinger, and L. L. Whitcomb, “Development and evaluation of anactuated MRI-compatible robotic system for MRI-guided prostate inter-vention,” IEEE/ASME Trans. Mechatronics, vol. 18, no. 1, pp. 273–284,Feb. 2013.

[8] S.-E. Song, J. Tokuda, K. Tuncali, C. Tempany, E. Zhang, and N. Hata,“Development and preliminary evaluation of a motorized needle guidetemplate for MRI-guided targeted prostate biopsy,” IEEE Trans. Biomed.Eng., vol. 60, no. 11, pp. 3019–3027, Nov. 2013.

[9] D. Stoianovici, D. Song, D. Petrisor, D. Ursu, D. Mazilu, M. Muntener,M. Mutener, M. Schar, and A. Patriciu, “MRI stealth robot for prostateinterventions,” Minimally Invasive Ther. Allied Technol., vol. 16, no. 4,pp. 241–248, 2007.

[10] D. Stoianovici, C. Kim, G. Srimathveeravalli, P. Sebrecht, D. Petrisor, J.Coleman, S. Solomon, and H. Hricak, “MRI-safe robot for endorectalprostate biopsy,” IEEE/ASME Trans. Mechatronics, vol. 19, no. 4,pp. 1289–1299, Aug. 2014.

[11] G. S. Fischer, I. Iordachita, C. Csoma, J. Tokuda, S. P. DiMaio, C. M.Tempany, N. Hata, and G. Fichtinger, “MRI-compatible pneumatic robotfor transperineal prostate needle placement,” IEEE/ASME Trans. Mecha-tronics, vol. 13, no. 3, pp. 295–305, Jun. 2008.

[12] R. Seifabadi, S.-E. Song, A. Krieger, N. B. Cho, J. Tokuda, G. Fichtinger,and I. Iordachita, “Robotic system for MRI-guided prostate biopsy: Fea-sibility of teleoperated needle insertion and ex vivo phantom study,” Int.J. Comput. Aided Radiol. Surg., vol. 7, pp. 181–190, 2011.

SU et al.: PIEZOELECTRICALLY ACTUATED ROBOTIC SYSTEM FOR MRI-GUIDED PROSTATE PERCUTANEOUS THERAPY 1931

[13] R. Gassert, A. Yamamoto, D. Chapuis, L. Dovat, H. Bleuler, and E. Burdet,“Actuation methods for applications in MR environments,” ConceptsMagn. Reson. Part B, Magn. Reson. Eng., vol. 29B, no. 4, pp. 191–209,2006.

[14] D. Yakar, M. G. Schouten, D. G. H. Bosboom, J. O. Barentsz, T. W. J.Scheenen, and J. J. Futterer, “Feasibility of a pneumatically actuated MR-compatible robot for transrectal prostate biopsy guidance,” Radiology,vol. 260, no. 1, pp. 241–247, 2011.

[15] B. Yang, U.-X. Tan, A. B. McMillan, R. Gullapalli, and J. P. Desai,“Design and control of a 1-DOF MRI-compatible pneumatically actuatedrobot with long transmission lines,” IEEE/ASME Trans. Mechatronics,vol. 16, no. 6, pp. 1040–1048, Sep. 2011.

[16] Y. Chen, C. Mershon, and Z. Tse, “A 10-mm MR-conditional unidirec-tional pneumatic stepper motor,” IEEE/ASME Trans. Mechatronics, vol.PP, no. 99, pp. 1–7, 2014.

[17] J. Huber, N. Fleck, and M. Ashby, “The selection of mechanical actuatorsbased on performance indices,” in Proc. Royal Soc. London. Ser. A, Math.,Phys. Eng. Sci., vol. 453, no. 1965, pp. 2185–2205, 1997.

[18] D. Stoianovici, A. Patriciu, D. Petrisor, D. Mazilu, and L. Kavoussi, “Anew type of motor: Pneumatic step motor,” IEEE/ASME Trans. Mecha-tronics, vol. 12, no. 1, pp. 98–106, Feb. 2007.

[19] G. S. Fischer, A. Krieger, I. Iordachita, C. Csoma, L. L. Whitcomb,and G. Fichtinger, “MRI compatibility of robot actuation techniques—A comparative study,” in Proc. Int. Conf. Med. Image Comput. Comput.Assisted Intervention., Sep. 2008.

[20] H. Elhawary, Z. T. H. Tse, A. Hamed, M. Rea, B. L. Davies, andM. U. Lamperth, “The case for MR-compatible robotics: A review of thestate of the art,” Int. J. Med. Robot. Comput. Assisted Surg., vol. 4, no. 2,pp. 105–113, 2008.

[21] J. J. Futterer, S. Misra, and K. J. Macura, “MRI of the prostate: Potentialrole of robots,” Imaging, vol. 2, no. 5, pp. 583–592, 2010.

[22] A. Goldenberg, J. Trachtenberg, W. Kucharczyk, Y. Yi, M. Haider, L. Ma,R. Weersink, and C. Raoufi, “Robotic system for closed-bore MRI-guidedprostatic interventions,” IEEE/ASME Trans. Mechatronics, vol. 13, no. 3,pp. 374–379, Jun. 2008.

[23] H. Elhawary, Z. Tse, M. Rea, A. Zivanovic, B. Davies, C. Besant,N. de Souza, D. McRobbie, I. Young, and M. Lamperth, “Roboticsystem for transrectal biopsy of the prostate: Real-time guidance un-der MRI,” IEEE Eng. Med. Biol. Mag., vol. 29, no. 2, pp. 78 –86,Mar. 2010.

[24] G. S. Fischer, G. A. Cole, and H. Su, “Approaches to creating and con-trolling motion in MRI,” in Proc. IEEE Annu. Int. Conf. Eng. Med. Biol.Soc., 2011, pp. 6687–6690.

[25] J. Tokuda, G. S. Fischer, X. Papademetris, Z. Yaniv, L. Ibanez, P. Cheng,H. Liu, J. Blevins, J. Arata, A. J. Golby, T. Kapur, S. Pieper,E. C. Burdette, G. Fichtinger, C. M. Tempany, and N. Hata,“OpenIGTLink: An open network protocol for image-guided therapy en-vironment,” Int. J. Med. Robot. Comput. Assisted Surg., vol. 5, no. 4,pp. 423–434, 2009.

[26] H. Su, M. Zervas, G. Cole, C. Furlong, and G. S. Fischer, “Real-time MRI-guided needle placement robot with integrated fiber optic force sensing,”in Proc.IEEE Int. Conf.Robot. Autom, 2011, pp. 1583–1588.

[27] Y. Yu, T. Podder, Y. Zhang, W. Ng, V. Misic, J. Sherman, L. Fu,D. Fuller, E. Messing, D. Rubens, J. Strang, and R. Brasacchio, “Robot-assisted prostate brachytherapy,” Med. Image Comput. Comput.-AssistedIntervention, vol. 9, pp. 41–49, 2006.

[28] H. Su, D. Cardona, W. Shang, A. Camilo, G. Cole, D. Rucker, R. Webster,and G. Fischer, “MRI-guided concentric tube continuum robot with piezo-electric actuation: A feasibility study,” in Proc. IEEE Int. Conf. Robot.Autom., 2012, pp. 1939–1945.

[29] W. Shang and G. S. Fischer, “A high accuracy multi-image registra-tion method for tracking MRI-guided robots,” Proc. SPIE Med. Imag.,vol. 8316, 2012.

[30] D. Stoianovici, “Multi-imager compatible actuation principles in surgi-cal robotics.” Int. J. Med. Robot. Comput. Assisted Surg., vol. 1, no. 2,pp. 86–100, Jan. 2005.

[31] Determination of Signal-to-Noise Ratio (SNR) in Diagnostic MagneticResonance Imaging, NEMA Standard Publication MS 1-2008, 2008.

[32] H. Su, W. Shang, K. Harrington, A. Camilo, G. A. Cole, J. Tokuda,C. Tempany, N. Hata, and G. S. Fischer. (2012), “A networked modularhardware and software system for MRI-guided robotic prostate interven-tions,” Proc. SPIE Med. Imag., vol. 8316, [online]. http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1285684.

Hao Su (M’12) received the B.S. degree from theHarbin Institute of Technology and the M.S. de-gree in mechanical and aerospace engineering fromthe State University of New York, Buffalo, NY, USA,and received the Ph.D. degree in mechanical en-gineering from the Worcester Polytechnic Institute,Worcester, MA, USA.

He is currently a Research Scientist at Philips Re-search North America, Briarcliff Manor, NY, USA.

Dr. Su received the Best Medical Robotics PaperFinalist Award in the IEEE International Conference

on Robotics and Automation. He received the Link Foundation Fellowship andDr. Richard Schlesinger Award from the American Society for Quality.

Weijian Shang (S’12) received the B.S. degree inprecision instruments and mechanology from Ts-inghua University, Beijing, China, in 2009, and theM.S. degree in mechanical engineering from Worces-ter Polytechnic Institute, Worcester, MA, USA, in2012, where he is currently working toward the Ph.D.degree in mechanical engineering.

Gregory Cole (M’12) received the B.S. and M.S.degrees in mechanical engineering and the Ph.D. de-gree in robotic engineering in 2013 from WorcesterPolytechnic Institute, Worcester, MA, USA.

He is currently a Research Scientist at Philips Re-search North America, Briarcliff Manor, NY, USA.

Gang Li (S’12) received the B.S. and M.S. degreesin mechanical engineering from the Harbin Instituteof Technology, Harbin, China, in 2008 and 2011,respectively. He is currently working toward the Doc-toral degree in the Department of Mechanical Engi-neering at Worcester Polytechnic Institute, Worcester,MA, USA.

Kevin Harrington received the B.S. degree inrobotic engineering from Worcester Polytechnic In-stitute, Worcester, MA, USA.

He is currently with the Automation and In-terventional Medicine Robotics Lab, Departmentof Mechanical Engineering, Worcester PolytechnicInstitute.

Alexander Camilo received the B.S. degree in elec-trical and computer engineering from WorcesterPolytechnic Institute, Worcester, MA, USA.

He is currently with the Automation and In-terventional Medicine Robotics Lab, Departmentof Mechanical Engineering, Worcester PolytechnicInstitute.

1932 IEEE/ASME TRANSACTIONS ON MECHATRONICS, VOL. 20, NO. 4, AUGUST 2015

Junichi Tokuda received the B.S. degree in engi-neering, the M.S. and Ph.D degrees in informationscience and technology in 2002, 2004, and 2007, re-spectively, all from the University of Tokyo, Tokyo,Japan.

He is currently an Assistant Professor of Radiol-ogy at Brigham and Women’s Hospital and HarvardMedical School, Boston, MA, USA.

Clare M. Tempany received the MB BAO, BCh de-grees from the Royal College of Surgeons, Dublin,Ireland.

She is currently the Ferenc Jolesz Chair and theVice-Chair of Radiology Research, Department ofRadiology, Brigham and Women’s Hospital, Boston,MA, USA, and a Professor of Radiology at HarvardMedical School, Boston. She is also the Clinical Di-rector of the National Image Guided Therapy Center.

Nobuhiko Hata received the B.E., M.E., and D.Eng.degrees in precision machinery engineering in 1993,1995, and 1998, respectively, all from the Universityof Tokyo, Tokyo, Japan.

He is currently an Associate Professor of radiologyat Harvard Medical School, Boston, MA, USA.

Gregory S. Fischer (M’03) received the B.S. degreein electrical and mechanical engineering from Rens-selaer Polytechnic Institute, Troy, NY, USA, and theM.S.E. degree in electrical engineering and the Ph.D.degree in mechanical engineering in 2008 from JohnsHopkins University, Baltimore, MD, USA.

He is currently an Associate Professor of me-chanical engineering at Worcester Polytechnic Insti-tute, Worcester, MA, USA, and is the Director of theWPI Automation and Interventional Medicine (AIM)Laboratory.